Implementing an original approach based on a virological study using home-based blood self-sampling which was nested in a large phone-based survey, our work provides new estimates for the public health burden of CHC and CHB in the general adult population living in mainland France: 133,466 persons (95% CI: 56,880-312,626) had CHC, among whom 80.6% (95% CI: 44.2-95.6) were aware of their infection; 135,706 persons (95% CI: 58,224-313,960) had CHB, among whom only 17.5% (95% CI: 4.9-46.4) were aware of their infection. Our work also provides useful data on HCV, HBV and HIV screening history which can guide the ongoing reassessment of current screening strategies. More specifically, approximately one in five and one in three persons reported HCV and HBV screening during their lifetime, respectively. If a universal (i.e., for all adults at least once in their life) combined (i.e. simultaneous) HCV/HBV/HIV screening strategy were implemented, between 33 and 85% of mainland France’s adult population would be concerned (i.e., between 15 and 40 million people).
To date, the only national survey-based prevalence data for HCV and HBV in France have come from a stand-alone survey conducted in 2004 on 14,500 individuals in social security medical centers [5]. The large sample size of the 2016-Health Barometer survey (2016-HB) together with its innovative sampling design [23], provided an excellent opportunity to implement an innovative nested survey (BaroTest) to collect hepatitis prevalence data while optimizing human and financial resources. Like BaroTest, other studies have also used blood self-sampling. However, they mainly focused on populations at high risk of HIV infection (MSM, black Africans) [31, 32] or sick people [33]. We found only one previous study investigating blood self-sampling in the general population, specifically in women participating in the Norwegian Breast Cancer Screening program [34]. To our knowledge, BaroTest is the first survey on HCV, HBV and HIV screening based on home-based blood self-sampling among the general population. A comprehensive system of reminders was put in place to prompt those who had agreed to participate to carry out the home-sampling (Additional file 1: Figure S1) [22]. Of all those initially invited to participate in the survey, almost 40% returned DBS, demonstrating that home-based blood self-sampling may constitute an effective and inexpensive tool in epidemiological studies.
Thanks to this relatively high participation rate and the large sample size in 2016-HB, almost 7,000 DBS (substantially more than the 5,000 expected) were analyzed, allowing us to estimate prevalence for CHC and CHB despite suboptimal power. We estimated the prevalence of CHC at 0.30% (95% CI: 0.13-0.70) and 0.30% (95% CI: 0.13-0.70) for CHB, in the general adult population in mainland France. In 2004, the national prevalence survey provided estimates of 0.53% (95% CI: 0.40–0.70) and 0.65% (95% CI: 0.45–0.93), respectively [5]. Since 2004, estimates for CHC prevalence in the general population in mainland France have been based on models: 0.42% (95% uncertainty interval: 0.33-0.53) for 2011, and 0.3% (95% uncertainty interval (UI): 0.1-0.3) and 0.29% (95% UI: 0.14-0.34) both for 2015 [4, 16, 17]. For CHB, the only estimate since 2004 has also been model based: 0.5% (95% UI: 0.4-0.7) for 2016 [15]. Due to methodological differences, caution is needed when comparing these estimates. In particular, compared with the 2004 prevalence survey based on venous blood sampling, the use of DBS in BaroTest may have led to a slight underestimation of CHC and CHB prevalences due to a possible lack of sensitivity. However, several studies, including meta-analyses, have shown excellent diagnostic accuracy (with both specificity and sensitivity higher than 98%) using DBS compared with venous blood sampling for the detection of anti-HCV antibodies, HBsAg and HCV RNA [24, 35, 36]. Although not significant, the observed decrease in CHC prevalence in the general population in mainland France may be linked to the decrease in the number of people infected by blood transfusion before 1992, the availability of DAAs, and the almost certain reduction in HCV incidence. The latter point is partly based on the very probable decrease in HCV incidence among drug users, suggested by a reduction in the prevalence of anti-HCV antibodies among this population (from 58.2% in 2004 to 43.2% in 2011) [37]. This decrease coincides with the continuous enhancement of harm reduction measures. In addition, with systematic screening of blood donations since 1992, the risk of transfusion-transmitted HCV infection is now extremely low, estimated at 0.03 per million donations in 2014-2016 [38]. With regard to CHB, the prevalence estimate from BaroTest did not significantly differ from previous estimates [5, 15]. Both our CHC and CHB prevalence estimates are in line with recent estimates in western European countries: 0.2- 0.25% for Germany [17, 39], 0.29% for the UK [17] for CHC; 0.3% for Germany [15, 39], 0.7% for the UK, 0.1% for Ireland [15] for CHB.
As 2016-HB focused on infectious diseases and sexual health, we were able to document the main HCV and HBV risk exposure factors. In the univariate analysis, CHC and CHB prevalences were significantly higher in individuals with well-known risk exposure factors, e.g. intravenous (IV) drug use during lifetime for CHC and being born in Sub-Saharan Africa for CHB. This finding strengthens the validity of our results, although multivariate analysis could not be performed due to the low number of individuals testing positive in BaroTest. Furthermore, CHC and CHB were significantly more frequent among people with a low socio-economic status, as previously demonstrated in the 2004 French national prevalence survey [5]. The absence of a significant difference in CHC prevalence between persons with a history of blood transfusion before 1992 and those without such a history may be explained by a lack of power and possibly by a memory bias. This suboptimal power may also contribute to CHB prevalence not being higher in men than in women. Caution is needed when extrapolating our results to the French general population. Indeed, marginalized populations where CHC or CHB prevalence is likely to be higher were either not represented at all because of the eligibility criteria (e.g., non-French speaking migrants), or probably underrepresented (e.g. drug users, homeless people) given that HB-2016 recruitment was carried out by telephone. This probable underestimation is particularly true for IV drug users as although some BaroTest participants reported IV drug use in their lifetime, this may have referred to past activity. Consequently, CHC prevalence among them (12%) is much lower than among active IV drug users recruited in harm reduction centres (30%) [40].
Among individuals testing positive, an estimated 80.6% were aware of their infection for CHC, but only 17.5% for CHB. These estimates are not robust given the very small numbers of persons testing positive. Consequently, the confidence intervals were large. In addition, given the previously described low level of knowledge regarding viral hepatitis among the French general population [41], questions in the survey regarding screening history could not specify any virological marker. This may have led to an overestimation of the proportion of infected individuals aware of their chronic infection, in particular for HCV. Possible participant confusion between the various types of hepatitis may also have impacted estimations of the percentages of people reporting lifetime screening for one disease or the other. Accordingly, careful interpretation of these values is needed, in particular when creating parallels with the proportions of people aware of their infection. Compared with previous estimates using the same methodology, the proportion of people reporting a history of testing is very close for HCV (19.2% in 2016-HB vs. 19.7% in 2010-HB [42]), but markedly differs for HBV (35.6% in 2016-HB vs. 15.2% in 2010-HB [42] and 27.4% in the 2010-KABP survey [41]). However, in multivariate analysis, factors associated with HCV or HBV screening history were consistent with those previously described, in particular regarding age (with screening more frequently reported by the 31-45 years old group and less often by the oldest group), educational level and level of urbanization of their place of residence [41, 42]. One important result is that after adjustment, HCV and HBV screening were significantly more frequent in individuals reporting risk exposure factors (e.g., IV drug use, being born in an endemic country). These groups constitute the target populations for the current French screening strategy. Having said that, the level of screening is insufficient (e.g., 65% for HCV screening among persons reporting IV drug use, 50% for HBV screening among migrants born in HBV highly endemic countries).
If the proposed universal combined HCV/HBV/HIV screening strategy were recommended, a large proportion of the general population in mainland France would be concerned. Indeed, based on declarative data, an estimated 85% of the general population aged 15-75 years have never been tested for all three diseases (33% have never been tested for any of the three). Given the probable poor reliability of self-reported HCV and HBV statuses [43, 44], this estimate most likely does not reflect the true proportion of individuals who have not been tested for all three viruses. However, if the HAS recommends this strategy, its practical implementation by physicians will rely on patients’ self-reported history of HCV/HBV/HIV screening. Therefore, this estimate is essential when implementing cost-effectiveness analyses. Irrespective of the strategy finally chosen, home blood self-sampling could prove to be an additional tool to enhance screening. Detailed data on its acceptability and feasibility in the context of BaroTest will be the subject of a future article.